Xinyi Li
Clemson University, SC·Mathematics & Statistics
Doctors now collect many kinds of data about each patient, including brain scans, genetic profiles and routine clinical measurements. These data are rarely combined in ways clinicians can trust and understand when choosing treatments.
- Fields
Mathematics & Statistics
Computer Science
Public Health & Epidemiology
- Topics
precision medicine
individualized treatment rules
dynamic treatment regimes
- Grants
- 1 known grant, through Aug 2029
What they work on
- Develop interpretable statistical learning methods for single-stage individualized treatment rules that combine medical imaging, genomics and clinical variables, with valid uncertainty quantification and statistical inference.
- Develop interpretable direct learning, or policy search, that estimates the best treatment rule without explicitly modeling outcomes, to improve robustness on large, noisy multimodal data.
- Develop interpretable reinforcement learning for dynamic treatment regimes that adapt to a patient's changing status, including longitudinal biomarkers, with theoretical guarantees such as large-sample properties.
- Evaluate the framework on real-world data from the Alzheimer's Disease Neuroimaging Initiative (ADNI) for early detection and risk stratification, and release scalable open-source software.
More about the research
This collaborative project develops transparent statistical and artificial intelligence (AI) methods that merge these data types into clear treatment guidance. That guidance should be tailored to each patient, adapt as the patient's condition changes and come with measurable reliability.
Why it matters
The methods could help clinicians use imaging, genetic and clinical data together to decide on treatment, with recommendations that can be explained and whose reliability is measured. Demonstrations on Alzheimer's disease data target earlier detection, risk stratification and more personalized treatment. Free open-source software could make these tools easier for other researchers to adopt and reproduce.
Known grants
From NIH and NSF research grants. Labs often have other funding we can't see.
Closest labs
Yang Ni
University of Texas at Austin, TX·Mathematics & Statistics
Develop a Bayesian joint model of treatment, outcome, covariates, and hidden confounders that identifies causal effects from observational data without instrumental or proxy variables.
Kaizheng Wang
Columbia University, NY·Mathematics & Statistics·Early-career
Develop methods to measure the overall fidelity of black-box generative AI models using effective sample size, an interpretable estimate of how much reliable information synthetic data contain.
Caleb Miles
Columbia University, NY·Mathematics & Statistics
Develop simple parametric and semiparametric efficient estimators of the average treatment effect in multi-study, multi-outcome data fusion, using related follow-up measurements as proxies for outcomes missing in some trials.
The school
Clemson University: 19 labs on PI Pages, $339M a year on research, 297 research doctorates a year.
Reach out
A start for your email. Fill in the brackets.